workplace-based vocational training in central and eastern europe - evidence on the school to work...
TRANSCRIPT
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
1/36
Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to
Work Transition of Hungarian Apprentices
Daniel Horn
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
2/36
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
3/36
Daniel Horn
Center for Economic and Regional Studies, HAS and ELTEcon
WP4 Task 3P.4.8.
Versionfinal 2013. 10. 29.
Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to
Work Transition of Hungarian Apprentices1
Abstract:
Although apprenticeship training has been praised for its effectiveness in smoothing the school-
to-work transition of non-college bound students in Western European dual education systems,
there is a lack of evidence from Central Eastern Europe. Using a unique individual-level panel
database, which includes an extensive set of controls, the study shows that Hungarian students of
the non-college bound vocational training track with workplace-based training, have about 10-15%
higher probability of initial employment, compared to similar graduates from the same track, who
were trained in school. This effect seems to be stable across industries, and robust to specification
checks.
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
4/36
1. Introduction2There is a serious lack of studies on the effectiveness of workplace-based vocational training in
Central-Eastern Europe (but see Baranowska, Gebel, and Kotowska 2011; Noelke and Horn 2011).
Although, it has been widely argued that workplace-based training helps preparing non-college
bound youth for the labor market, the arguments are almost solely based on studies from Western
Europe, mostly from dual vocational education and training (VET) systems. These studies show that
secondary level education systems, combining school-based vocational education with employer-
provided, workplace-based (apprentice) training, have sustained a positive track record in smoothing
the school to work transition process, lowering the unemployment rate, and increasing the quality of
work (Rosenbaum et al. 1990; Mller and Shavit 1998; Shavit and Mller 2000; Ryan 2001; Breen
2005; Wolbers 2007; Wolter and Ryan 2011; Piopiunik and Ryan 2012). Besides the lack of empirics
from Central-Eastern Europe, existing empirical research provides little information about the causaleffect of apprentice training on school-to-work transition.
The paper improves on the existing literature in these aspects. Firstly, it provides information on the
effectiveness of apprentice training from a non-dual education system in a Central-Eastern European
country, Hungary. Secondly, there has been only a handful of studies that tried to show the causal
effect of workplace-based training on labor market outcomes (Bonnal, Mendes, and Sofer 2002;
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
5/36
socialist era as well, but its essence has changed dramatically, and has much less resemblance today
to the traditional dual systems.
The nature of apprenticeship training in socialist countries was changed mainly due to the huge
national enterprises, as compared to the pre-war network of smaller independent firms and
businesses, where apprenticeship training had originated. The provision as well as the content of
training changed markedly. Also since the competitive market was demolished, and people were
essentially guaranteed a job, having an early work-experience mattered much less both for the
individual and the firm, which decreased the motivational aspects for both parties (see West 2013 for
a comprehensive review of the transition of the CEE VET systems). Under the socialist era the
students spent most their practical training time in large enterprises, rather than being in contract
with several smaller factories, but at the same time not being jointly educated in shared off-the-job
facilities, as in the dual apprentice training systems (Kogan 2008)
Moreover, although the post-transition changes in these countries have seen the demolishment of
the national enterprises and the revival of small businesses, the development of the education
system was highly decentralized (Horn 2010), and thus the central coordination of vocational
workplace-based training between business and the state was secondary, at best. In the beginning
of the privatization and restructuring processes in CEE countries [], employers largely withdrew
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
6/36
that look at Central and Eastern Europe. Also getting comparable statistics is difficult due to the
different understanding of apprenticeship training: many CEE countries refer to their vocational
training programs as apprentice training programs, although students dont have any contractual
agreements with firms (Kogan 2008, 2122), or label non-formal forms of training as apprentice
training3.
To my best knowledge there have been only two comprehensive reviews on the vocational education
sector in Central and Eastern Europe. Kogan, Gebel and Noelke (2008) have relied on local expertise
for their handbook to describe the general education system, including its VET, of the given
country. With this local help they have categorized the countries into school-based or dual VET
(see table 1 below).
Table 1. The predominant form of organization of VET
School-based Dual system
Bulgaria X
Czech Republic X
Estonia X
Hungary X
Latvia X
Lith i X
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
7/36
52). Thus it seems from these two sources that apprenticeship training is still significant in the
previous Austro-Hungarian empire countries and in Poland (with strong linkages to Germany), even
though apprenticeship training (or dual-education) might not mean the same as in Western Europe.
Looking at official statistics is less helpful due to the lack of comparable education data on
apprenticeship training, and due to the different understanding of the term. On the other hand, one
might approach training from the other direction, from the business sector. Table 2 below shows the
training enterprises as percentage of all enterprises. Apparently the typical dual education systems of
Austria, Denmark or Germany are all above the EU mean. But the conclusion from this table is less
straightforward. Firstly, this statistic is also high in non-dual Western countries (as the UK or Sweden
and Finland), and secondly, countries from Central and Eastern Europe are scattered along the
distribution (although on average below the EU mean). Note, however, that with the exception of
Estonia all ex-Austro-Hungarian empire members have higher percentages of training firms than
the other CEE countries. This also points toward the fact that workplace-based training can be
important in the CEE countries, and especially in those, where dual-training have a tradition. But
whether apprentice training is just as effective in this region as it is in the Western dual economies is
yet unclear.
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
8/36
Table 2. Training enterprises as % of all enterprises
2005 2010
United Kingdom 90 80
Norway 86 n.a.
Denmark 85 n.a.
Austria 81 87
Sweden 78 87
Finland 77 74
Netherlands 75 79
France 74 76
Slovenia 73 68
Czech Republic 72 72
Luxembourg 72 71
Germany (until 1990 former territory of the FRG) 69 73
Estonia 67 68
Ireland 67 n.a.
Belgium 63 78Slovakia 60 69
European Union (27 countries) 60 66
Croatia n.a. 57
Cyprus 51 72
Hungary 49 49
Spain 47 75
Lith i 46 52
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
9/36
In Hungary every student in the lowest non-college bound vocational training track had to do at
least two years of practical training, which could either be done in the school or at a firm (see the
description of the Hungarian VET system below).5This study compares these two groups of students.
Hence the treatment and the control groups are quite obvious: both groups have received
exactly the same general training (the first two years in the vocational training program), and
provided they aim for the same occupational qualification they have to fulfill the same
requirements; the only difference between the groups is the place where practical training takes
place. Although the allocation of students between training places might not be random, the HLCS
offers an exceptionally wide variety of individual controls, which reduces the omitted variable bias
concern. The database also includes information on the types of qualification that students have
acquired, and on the schools of the students, which allows for within industry and within school
analysis, that further strengthens the reliability of the estimates. Moreover, the HLCS is a panel
database, which rules out the problem of reverse causality.
4. Previous research on causal effectsThere are but a handful of empirical studies that offer analysis of the causal effects of apprenticeship
t i i i di id l l l l b k t t ( i b W lt d R 2011) Th
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
10/36
The authors argue that this finding could be due to three distinct reasons, among which they are
unable to discriminate: a) apprentices might lack the general human capital, as opposed to non-
apprentice VET students, and thus finding a job at a firm other than their training firm is harder/not-
easier; b) movers might be negatively selected, as those who are not hired by the training firm
might have some unobserved negative trait; and similarly c) there might be a negative signaling
effect associated with moving to another firm, even if movers are not different from stayers in
other respects.
A similarly designed study is Bertschy, Cattaneo and Wolter (2009), who look at the Swiss dual
system. They also use a panel which is connected to the PISA 2000 Swiss database, which provides
standardized test scores to proxy student achievement as well as socio-economic status and other
controls, and they also use simultaneously estimated equations to take selection into account.
However, since the vast majority of the Swiss vocational students (over 90%) are in the dual
apprenticeship training, they compare apprentices, who taken up training with higher intellectual
level, with the others. Also their utilized outcome is not employment, but employment in adequate
job that matches the graduates qualifications. Initially they find a significant difference between
these two groups, which disappears after they take selection into tracks into account. They
emphasize that self-selection into educational tracks is very important. In fact, students with higher
PISA lit l lik l t d t d lik l t ll i ti l fi ld ith
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
11/36
5. A non-dual CEE system - the Hungarian VET systemWhile most of the studies that have addressed the question of the effectiveness of apprenticeship
training are based on countries with dual systems, the Hungarian VET also allows for a within track
comparison of workplace-based and school-based training. As a non-dual system this country study
should especially be important for countries with less experience in apprentice training. Thus findings
from a non-dual system, where the workplace-based training is still widely utilized, could be
informative for those countries where apprenticeship training is less widely spread but its
development is considered.
Also, as van de Werfhorst and co-authors (van de Werfhorst 2011b; Bol and van de Werfhorst 2011;
van de Werfhorst 2011a) have pointed out different theories might explain better the education-
labor market link in different countries or in different labor market settings. Specifically van de
Werfhorst (2011b) argues that in dual systems the human capital theory is more adequate, since
there is a stronger match between the skills acquired in education and skills needed on the
workplace. In countries with less evident link between labor market and education, other indicators
of skills such as general literacy or numeracy are more important, and thus educational
attainment is less important. Although in his comparative studies (van de Werfhorst 2011b; Bol and
van de Werfhorst 2011) Hungary is considered as a strongly vocationally oriented country, and as
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
12/36
repeaters, or those students with special educational needs (SEN), who cannot be integrated with
the others and thus enrolled in special vocational training. While both the academic and the
vocational secondary tracks offers general training for four years - and the vocational secondary
offers pre-vocational training, with usually one or two optional years of vocational practical training
after the school-leaving exam the vocational training track has so far offered only two years of
general training7with two additional years of practical training.
This paper focuses on the 20%, who are enrolled in the vocational training (VT) track. This track is
considered to be the lowest ranked in the hierarchy of tracks (but still above no-education). This
paper compares VT students who have done practical training at a private firm with those, who have
done practical training within schools.
Although the selection into workplace-based training might not be random, there is no central
procedure that allocates students in one group or in another. In fact, the organization of the system
is overly school-based, with relatively few links to the labor market (Kis et al. 2008). The system has
been one of the most decentralized ones in the OECD (OECD 2004). It is the duty of the school to
provide practical training for the student. The school can either organize the training within its
boundaries (e.g. by hiring vocational teachers) or can outsource the training to a private firm,
which can be done in groups or individually as well. The student can also organize training for
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
13/36
between workplace-based and school-based training. After the introduction of the HLCS data, I will
address this question.
6. The HLCS dataThe Hungarian Life Course Survey (HLCS) is an individual panel survey conducted annually. The
original sample of 10,022 respondents was chosen in 2006 from the population of 108,932 eighth
grade students with valid test scores from the National Assessment of Basic Competencies (NABC).
The NABC measures the literacy and numeracy of all 6th
, 8th
and 10th
grade students every year,
starting from 2006 (OECD 2010). The NABC also contains a set of family background variables, such
as parental education or employment status. The first HLCS survey wave was completed during the
winter of the school-year 2006/7, and subsequent waves have been fielded on a yearly basis.
Currently there are 6 waves available with fairly large response rates. The annual sample attrition
rate, on average, is only around 5% (see Table 3).
Table 3. Basic statistics of the HLCS database
wave School Date of
th
Median school grade Number of students (with
li SEN
Number of students
( t ti b
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
14/36
students with special educational needs (SEN) are overrepresented in the data, propensity weights
are used to control for the oversampling, as well as for the imminent sample attrition. The following
strata were used during the data collection, and in estimating the weights: 1) 3 settlement types: the
capital and big cities, other cities, villages 2) 7 NUTS-2 regions93) Reading literacy test scores (30
equal groups from the NABC 2006 reading literacy distribution).
The most important variables of interest in this paper are the school track, the apprenticeship status,
and the labor market outcome. School track is defined as the students school track in the 4 thwave of
the study, the year when the median student was finishing the last year of compulsory schooling. All
students in the analysis were enrolled in the vocational training track in the 4th
wave. Vocational
training students could either do their practical training within school in class, or in a school
workshop, or could go to a private firm, either with the help of the school (usually in groups), or by
organizing the training by themselves. I have labeled the former two as school-based and the latter
two as workplace-based training. Anyone, who did workplace-based training in the 4th
wave or in the
5th
wave of the study (the year after finishing compulsory education), is considered an apprentice.10
The four types of labor market outcomes employed, unemployed, studying and other11 are
considered in the last (available) wave of the study, and are self-declared. The main reason for using
the 6th
wave and not the 5th
wave, which is the one after the compulsory education ends, is that the
j i f d i h 5th ill i d i h i l i i
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
15/36
15
Table 4: Labor market outcomes in the 5th
and 6th
wave
5th wave 6th wave
work unempl. study other missing Total work unempl. study other missing Total
academic 70 54 1717 62 172 2075 187 95 1419 85 289 2075
% 3,37 2,6 82,75 2,99 8,29 100 9,01 4,58 68,39 4,1 13,93 100
voc.sec. 106 115 2037 62 158 2478 452 303 1219 161 343 2478
% 4,28 4,64 82,2 2,5 6,38 100 18,24 12,23 49,19 6,5 13,84 100
voc.tr. 148 189 958 62 114 1471 541 290 286 123 231 1471
% 10,06 12,85 65,13 4,21 7,75 100 36,78 19,71 19,44 8,36 15,7 100
spec.voc.tr. 23 34 191 12 26 286 60 45 108 25 48 286
% 8,04 11,89 66,78 4,2 9,09 100 20,98 15,73 37,76 8,74 16,78 100
missing 252 418 906 246 1890 3712 508 408 515 262 2019 3712% 6,79 11,26 24,41 6,63 50,92 100 13,69 10,99 13,87 7,06 54,39 100
Total 599 810 5809 444 2360 10022 1748 1141 3547 656 2930 10022
% 5,98 8,08 57,96 4,43 23,55 100 17,44 11,38 35,39 6,55 29,24 100
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
16/36
Other variables that are used are the standardized test score (literacy and numeracy) in 8th
grade,12
class mark averages (1- fail to 5- excellent) in 8th
and in 12th
grade, gender, SEN status, Roma
ethnicity, parental education and occupation. All control variables are from the first wave of the
study, unless otherwise noted. Additional controls are a proxy for grade repetition (whether the
student was in the 12th
grade in the 4th
wave of the study) and a proxy for motivation (whether
her/his 9th
grade school was her/his first choice). The size of the training firm (small: 1 to 12
employed, medium: 13 to 100 employed and large: over 100 employed) is also used in some
estimations. The month of survey is controlled in all estimations and is not shown. See Table 5 for
descriptive statistics.
Table 5: Descriptive statistics
data available for students in the 6
th
wave of HLCSVocational training students only
Variable obs. weighted obs. mean s.d. min. max.
apprentice, 4thor 5
thwave 1183 15048 0.60 0.49 0 1
math test score (std.), 8thgrade 1087 14180 -0.83 0.68 -2.74 2.10reading test score (std.), 8
tgrade 1217 15447 -0.92 0.68 -3.78 1.21
class mark average, 8tgrade 1170 14883 3.18 0.53 1 5
l k 12
th
d 1217 15447 3 32 0 58 2 5
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
17/36
students only 964 has this information in the dataset. The official list of OKJ qualifications contains 21
larger categories. I have grouped these into 6 broad categories (industries) in order to increase the
number of cases within each category, but still facilitate relevant comparison between the groups
(see Table A1. in the appendix).
Table 6.: Number and percentage of VT students in school- and workplace-based training by industry
Industry school-based work-based missing Total
social services 3 6 0 9
% 33,33 66,67 0 100
mechanics 108 112 4 224
% 48,21 50 1,79 100
industry 124 106 2 232
% 53,45 45,69 0,86 100
transport-environment 13 19 0 32
% 40,63 59,38 0 100
services 121 267 7 395
% 30,63 67,59 1,77 100
agriculture 43 29 0 72
% 59,72 40,28 0 100
missing 178 296 33 507
% 35,11 58,38 6,51 100
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
18/36
Covariates that are significant in the first estimation (Table 7 column 1) suggest that higher skilled
students are more likely to enter apprenticeship training. The within industry estimation (column 2)
as well as the within school estimation (column 3) do not show these strong skill differences between
apprentices and non-apprentices, suggesting that the (self)selection into occupations or more likely
into schools drives the results. That is, there might be some occupations and/or schools that attract
better students (see also Bertschy, Cattaneo, and Wolter 2009). The base (column 1) and the within
school estimations (column 3) also show that people with less educated parents are more likely to
have practical training at private firms. The results in the most restrictive within school and within
occupation model (column 4) however highlight that none of the individual traits matter if
occupational differences and school and/or local labor market effects are taken into account.
Individual traits explain only 4% of the variation of the base model, but industry effects add an
additional 4%, suggesting that students in different occupations have different chances of getting
into apprenticeship (see also table 6). The inclusion of school fixed effects in the model increases the
fit tremendously. The school fixed effect model explains is almost 50% of the variance, which is
further increased to 73% when industry fixed effects are added. This is of course not surprising, given
that the sample was not representative on the school level and thus there are on average less than 4
students per school in the sample, which further decreases when within industry effects are
id d i hi h l
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
19/36
Table 7: Selection into apprenticeship linear probability models
(1) (2) (3) (4)
class mark average, 8th grade 0.0914** 0.0321 0.0289 -0.0428
(0.0413) (0.0505) (0.0444) (0.0794)
class mark average, 12th grade -0.0124 -0.0118 0.0611 0.0925
(0.0334) (0.0396) (0.0377) (0.0626)
math test score (std.), 8th grade -0.0141 -0.00424 -0.0280 -0.0182
(0.0318) (0.0378) (0.0318) (0.0530)
reading test score (std.), 8th grade 0.0624** 0.0337 0.0263 -0.0380
(0.0308) (0.0406) (0.0338) (0.0652)parents' ed.: primary or below 0.0866* 0.0689 0.134*** 0.0927
(0.0482) (0.0625) (0.0513) (0.0912)
parents' ed.: secondary or higher 0.0151 0.0260 0.00147 0.0226
(0.0437) (0.0521) (0.0456) (0.0782)
father employed, 4th wave -0.0553 -0.0292 -0.0711 -0.105
(0.0437) (0.0526) (0.0487) (0.0877)
father unemployed, 4th wave -0.0270 -0.00995 -0.0469 0.00422
(0.0508) (0.0667) (0.0539) (0.0911)
SEN student 0.0294 0.0210 0.0795 0.0878
(0.103) (0.0935) (0.103) (0.172)
Roma -0.0429 -0.0581 -0.0223 0.103
(0.0647) (0.0862) (0.0655) (0.119)
9th grade track is first choice 0.0323 0.0775 0.0531 0.104
(0.0426) (0.0549) (0.0441) (0.0743)
12th grader in 4th wave 0.142*** 0.151*** 0.0481 0.0822
(0.0476) (0.0547) (0.0483) (0.0768)
f l 0 0159 0 0680 0 0230 0 00427
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
20/36
significant in all three estimations, and show, that those VT students who had carried out practical
training at a private firm, as opposed to doing practical training in school, have around 1.5-1.6 times
higher odds of being employed, as opposed to being unemployed. The size of this effect is
unchanged by any of the personal traits that are included in the model. On the other hand social
background, gender and grade repetition matters in getting a job. It seems that students with
employed fathers have much higher odds of being employed; whether this effect materializes
through socialization or though social networks is not obvious. Also men are more likely to be
employed and women are more likely to fall into the other category (e.g. maternity leave). Students,
who have not repeated grades until 12th
grade, are also more likely to be employed in the 6th
wave of
the study. Note, however, that none of the school achievement variables neither the standardized
test scores, nor the teacher given class marks seem to be relevant in employment, although
students with higher class marks are more likely to study than to be unemployed.
Table 9 adds further controls to the base model. Table 9 block 1 is the same as Table 8 block 3 to
facilitate comparison of models. Table 9 block 2 shows the same multinomial logit model with
industry fixed effects added,14
while dummies of the training firm size are used in block 3 instead of
the apprenticeship dummy. The main conclusion does not change even if these controls are added:
apprentices have significantly higher odds to be employed vs. being unemployed than those with
l h l b d l l h h h ff l h l
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
21/36
Table 8: Effects of apprenticeship training, base model - multinomial logit model, odds of being employed, studying or other wrt. being unemployed
(1) (2) (3the base model)
work study other work study other work study other
apprentice, 4thor 5thwave 1.489** 1.149 1.226 1.457* 0.975 1.179 1.648** 1.106 1.211
(0.264) (0.232) (0.319) (0.283) (0.216) (0.324) (0.335) (0.253) (0.401)
class mark average, 8th grade 1.186 1.367 1.373 1.210 1.347 1.384
(0.258) (0.325) (0.438) (0.265) (0.312) (0.534)
class mark average, 12th grade 1.136 1.636** 1.597* 1.056 1.618** 1.344
(0.197) (0.323) (0.386) (0.188) (0.320) (0.436)
math test score (std.), 8th grade 1.170 1.062 0.720 0.963 0.972 0.974
(0.221) (0.216) (0.184) (0.184) (0.203) (0.272)
reading test score (std.), 8th grade 0.724* 1.023 0.897 0.797 1.064 0.681
(0.128) (0.209) (0.233) (0.144) (0.217) (0.203)
parents' ed.: primary or below 0.624* 0.529** 0.789
(0.167) (0.157) (0.317)
parents' ed.: secondary or higher 0.985 1.243 1.534
(0.235) (0.341) (0.712)
father employed, 4th wave 1.841** 1.247 1.707(0.443) (0.343) (0.693)
father unemployed, 4th wave 0.927 0.685 0.839
(0.254) (0.216) (0.365)
SEN student 0.807 0.851 8.19e-08***
(0.413) (0.490) (5.52e-08)
Roma 0.877 1.071 3.538***
(0.289) (0.433) (1.525)
9th grade track is first choice 1.015 1.017 1.084
(0.231) (0.259) (0.406)
12th grader in 4th wave 1.851** 0.603* 0.730
(0.483) (0.161) (0.279)
female 0.539*** 0.987 10.18***
(0.124) (0.257) (4.456)
Constant 1.487*** 0.923 0.349*** 0.501 0.0786*** 0.0190*** 0.372 0.154* 0.00588***
(0.193) (0.137) (0.0640) (0.415) (0.0754) (0.0207) (0.328) (0.158) (0.00820)
Observations 1,183 1,183 1,183 1,025 1,025 1,025 964 964 964
Standard error in parentheses, *** p
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
22/36
Table 9: Effects of apprenticeship training, industry FE - multinomial logit model, odds of being employed, studying or other wrt. being unemployed
(1the base model) (2) (3)
VARIABLES work study other work study other work study other
apprentice, 4th
or 5th
wave 1.648** 1.106 1.211 1.826** 0.978 1.329
(0.335) (0.253) (0.401) (0.479) (0.275) (0.594)
apprentice firm size, small (1-12)+ 1.496 0.790 0.983
(0.477) (0.276) (0.637)
apprentice firm size, medium (13-100)+ 3.926*** 2.275* 5.083**
(1.663) (1.039) (3.246)
apprentice firm size, large (100+)+ 1.703 0.937 0.799
(0.597) (0.381) (0.573)
Constant 0.372 0.154* 0.00588*** 0.103 0.563 0.0572 0.0739* 0.457 0.0450
(0.328) (0.158) (0.00820) (0.153) (0.863) (0.148) (0.112) (0.705) (0.109)
Industry FE no no no yes yes yes yes yes yes
Observations 964 964 964 679 679 679 670 670 670
Standard error in parentheses, *** p
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
23/36
Table 10 shows the predicted probabilities and marginal effects of apprenticeship training. The
baseline uncontrolled average probability of being employed for a VT student in 2011 is 44%.
Apprentices, however have a 47.1% chance, while school-based trained students have a 39.6%
chance of being employed. The chances of being unemployed is the reverse: apprentices have a 21%
chance, while the others have a 26.5% chance. There are no differences in the uncontrolled average
baseline probabilities of the other two outcomes between the two groups (study: 24%, other: 9%).
Using the base model (Table 8 block 3) to predict the probabilities at the population means similar
but somewhat higher percentages are gained. The predicted a probability of being employed for
apprentices is 52.3%, while for the school-trained it is 41.4%; the marginal effect of being trained at a
private firm is thus 10.9 % at the mean. In other words, the average apprentice has about 11% higher
chance of being employed after graduation than the average non-apprentice. This effect is somewhat
lower for the top of the distribution students (high class mark averages, high literacy and numeracy,
and parents with secondary general or tertiary schooling) and higher for the lower status lower
skilled students (low class mark averages, low literacy and numeracy, and parents education primary
or below). While the former group has 8.7% higher probability of being employed the latter has
11.5%. The marginal effects are also larger for apprentices, who were trained in mid-sized firms. The
average marginal effect here is 19%, but bottom of the distribution students benefit more (25.7%)
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
24/36
Table 10: Predicted probabilities and marginal effects
Predicted probability Marginal effect
school-based
training
workplace-based
training
workplace-based training
at population mean atpopulati
on mean
For a lowachiever low
status
student*
For a highachiever, high
status
student**
uncontrolled 39.6 47.1 - - -
base model 41.4 52.3 10.9 11.5 8.7
with industry FE 35.6 50.4 14.8 14.9 13.4
at a medium size firm 34.9 53.9 19.0 25.7 14.5within industry
at population mean at population
mean
at industry mean
social services 13.9 23.0 9.4 14.6
mechanics 39.8 54.9 15.2 14.8
industry 36.7 51.6 15.0 15.2
transport-environment 44.9 60.3 15.3 15.3
services 33.7 48.2 14.7 15.3agriculture 28.0 41.5 13.8 15.3
* class mark averages =2, standardized test scores=-1, highest parental education= primary or below, non-SEN,
non-Roma, non-repeater, current track is first choice, male
** class mark averages =5, standardized test scores=1, highest parental education= secondary or above, non-
SEN, non-Roma, non-repeater, current track is first choice, male
9. Robustness checks
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
25/36
apprenticeship training, since it controls for both local labor market effects as well as potential
differences between school qualities. Note that the HLCS has not used schools as sampling units, thus
the fact that some students are from the same school is chance only. There are 16 schools with only
one student in the sample. The average school has 3.7 students in the sample, which further
decreases to 3.1 if industry fixed effects are included. Taking missing values as well as the variance of
the outcome measure within school into account, and the fact that a representative subsample
should be used due to problems of weighting in fixed-effects logistic regressions, little less than 100
schools would be left for a non-linear analysis.16
Also since the multinomial logit model with a large
number of fixed effects has not yet been fully developed (see Pforr 2011), linear probability models
were estimated for this robustness check.17
The four category outcome of employed, unemployed,
study or other was transformed into a binary as in the propensity score matching test (employed vs.
everyone else).
The effect of apprenticeship training halves within schools, and loses its significance if school fixed-
effects are included in the linear model (Table 11). Apparently, the average apprenticeship student
does not have a greater chance of being employed than the average non-apprentice if they went to
the same school. However this effect is driven by the size of the training firm. Apprentices, who were
trained in small or in large firms, have exactly the same probability to be employed as non-
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
26/36
26
Table 11: Robustness check with industry and school fixed effects linear probability models
(1) (2) (3) (4) (5) (6) (7) (8)
apprentice, 4th
or 5th
wave 0.0986*** 0.127*** 0.0560 0.0570
(0.0380) (0.0464) (0.0477) (0.0629)
apprentice firm size, small (1-12) 0.0744 0.0993* -0.0117 0.0170
(0.0477) (0.0567) (0.0576) (0.0729)apprentice firm size, medium (13-100) 0.182*** 0.183*** 0.190*** 0.156*
(0.0518) (0.0675) (0.0629) (0.0877)
apprentice firm size, large (100+) 0.0559 0.120* 0.0131 0.0467
(0.0544) (0.0642) (0.0696) (0.0842)
Constant 0.365** 0.0342 0.442** 0.110 0.363** -0.0176 0.441** 0.0614
(0.164) (0.236) (0.183) (0.259) (0.166) (0.236) (0.186) (0.268)
Observations 964 675 957 670 936 666 929 661
R-squared 0.093 0.105 0.440 0.498 0.099 0.109 0.451 0.498
Industry FE no yes no yes no yes no yes
School FE no no yes yes no no yes yes
Dependent variable is employed=1, unemployed, studying or other =0, all variables as in the base model (Table 8 block 3) are controlled for
Robust standard errors in parentheses; *** p
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
27/36
The third robustness check sheds a bit of light on the firm-size puzzle. Figures 1 and 2 use another set
of outcome variables. The HLCS also asks students about their employment status during the last
academic year. That is, students in the 6th
wave of the study, in 2012 spring, were asked whether
they had had any regular job during the months between September 2010 (the start of the previous
school year) and August 2011, and students in the 5th
wave were asked whether they had a regular
job between September 2009 and August 2010. The data is for each month in between. Figure 1
depicts the marginal effect of apprenticeship training for a male, non-Roma, non-SEN student with
average class marks and test scores, parents with vocational education, who has not repeated class
up until 12th
grade, and applied for his track in the first place in 9th
grade, filled out the survey in May
2012 and have qualification from the average industry. The dependent variable is 1, if the student
had a regular job and 0 otherwise. Note that the outcome in May 2012 is the variable that was used
as in the estimations above.
It seems that apprentices are much more likely to find a regular job right after the end of the school
year. The marginal effect of apprenticeship training increases dramatically after the end of school
during the summer months, and declines afterwards. This indicates that apprentice VT students have
a quicker transition to the labor market than the non-apprentice VT students, but also that their
advantage slowly evaporates.18
The initial effect is also quite sizeable: it is around 19% in August
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
28/36
Figure 1. Marginal effect of apprentice students having a regular job
end of school-year 2010/11
end of school-year 2009/10
-.1
0
.1
.2
.3
10 Apr10 Jun
10 Sept10 Dec
11 Mar11 Jun
11 Aug12 May
95% confidence intervalls
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
29/36
Whether this effect is due to the superior specific skills that apprentices gained while being trained at
the mid-sized firms, or due to the screening of these firms, is not entirely clear from these figures.
Note however that the screening argument would predict an immediate and large difference
between the groupsbecause training firms hire the best candidates right after the trainingwhich
effect should converge by time, as firms hire new employees. The human capital argument, on the
other hand, would suggest a steady but continuous increase in the gap, which should only fade away
after a good amount of time, when others also gain the specific skills necessary for employment. The
figures, thus, support the screening rather than the human capital arguments. Also it is likely that
small firms rely less on institutionalized screening processes, as they are more likely to use their
social networks to recruit apprentices and thus rely less on this probation period and more on
other information channels (e.g. take relatives or friends as apprentices in the first place).
Conversely, medium or large firms are more likely to have institutionalized mechanisms for selecting
apprentices and new employees. Moreover, turnover at a large firm is probably larger, at least in
sheer numbers, which suggests that the potential advantage that an apprentice can gain from being
selected early diminishes quickly as the firm hires new workforce. Nevertheless this analysis in itself
is not enough to see whether the screening or the human capital argument comes closer to reality,
19
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
30/36
they finish education, than those who had their practical training in schools and are otherwise similar
to the workplace-based group. The effect is net of individual skills, school attainment, parental
background, motivation, gender and ethnicity, and robust to the inclusion of industry fixed effects,
and for school fixed effects but only for students trained in mid-sized firms. Also results show that
the significant marginal effect of apprenticeship training declines rapidly for students trained at large
firms, while this decline is less marked in medium or small firm trained apprentices, suggesting that
large turnover could eliminate the positive effects of apprenticeship training more quickly.
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
31/36
References
Baranowska, Anna, Michael Gebel, and Irena E. Kotowska. 2011. The Role of Fixed-term Contracts atLabour Market Entry in Poland: Stepping Stones, Screening Devices, Traps or Search
Subsidies? Work, Employment & Society25 (4) (December 1): 777793.doi:10.1177/0950017011419705.
Bertschy, Kathrin, M. Alejandra Cattaneo, and Stefan C. Wolter. 2009. PISA and the Transition intothe Labour Market. LABOUR23: 111137. doi:10.1111/j.1467-9914.2008.00432.x.
Bol, Thijs, and Herman G. van de Werfhorst. 2011. Signals and Closure by Degrees: The EducationEffect Across 15 European Countries. Research in Social Stratification and Mobility29 (1)
(January): 119132. doi:10.1016/j.rssm.2010.12.002.Bonnal, Liliane, Sylvie Mendes, and Catherine Sofer. 2002. School-to-Work Transition:
Apprenticeship Versus Vocational School in France. International Journal of Manpower23(5): 426442. doi:10.1108/01437720210436046.
Breen, Richard. 2005. Explaining Cross-National Variation in Youth Unemployment: Market andInstitutional Factors. European Sociological Review21 (2): 125134.
CEDEFOP. 2011. Vocational Education and Training in Poland. Luxembourg: Publications Office ofthe European Union.
Eichhorst, Werner, Nria Rodrguez-Planas, Ricarda Schmidl, and Klaus F. Zimmermann. 2012. ARoadmap to Vocational Education and Training Systems Around the World. IZA DP(7110):46.
Horn, Dniel. 2010. The Political Background of Structural Changes in the Educational System ofHungary, 1985-1994. Center for Policy Studies Working Paper(2010/1): 45.
. 2013a. Diverging Performances: The Detrimental Effects of Early Educational Selection onEquality of Opportunity in Hungary. Research in Social Stratification and Mobility.
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
32/36
Pforr, Klaus. 2011. Implementation of a Multinomial Logit Model with Fixed Effects. German StataUsers Group Meetings 2011. Stata Users Group.
http://econpapers.repec.org/paper/bocdsug11/03.htm.Piopiunik, Marc, and Paul Ryan. 2012. Improving the Transition Between Education/training and theLabour Market: What Can We Learn from Various National Approaches? EENEE AnalyticalReport(13.).
Plug, Erik, and Wim Groot. 1998. Apprenticeship Versus Vocational Education: Exemplified by theDutch Situation. Unpublished Manuscript.
Rosenbaum, James E, Takehiko Kariya, Rick Settersten, and Tony Maier. 1990. Market and NetworkTheories of the Transition from High School to Work: Their Application to Industrialized
Societies.Annual Review of Sociology16: 263299.Ryan, Paul. 2001. The School-to-Work Transition: A Cross-National Perspective.Journal of
Economic Literature39 (1): 3492.Shavit, Yossi, and Walter Mller. 2000. Vocational Secondary Education: Where Diversion and
Where Safety Net? European Societies21 (1): 2950.Van de Werfhorst, Herman G. 2011a. Skills, Positional Good or Social Closure? The Role of Education
Across Structuralinstitutional Labour Market Settings.Journal of Education and Work24(5): 521548. doi:10.1080/13639080.2011.586994.
. 2011b. Skill and Education Effects on Earnings in 18 Countries: The Role of NationalEducational Institutions. Social Science Research40 (4) (July): 10781090.doi:10.1016/j.ssresearch.2011.03.004.
West, John. 2013. Vocational Education and Training in Eastern Europe: Transition and Influence.LLAKES Research Paper(41): 68.
Wolbers, Maarten H. J. 2007. Patterns of Labour MarketEntry: A Comparative Perspective onSchool-to-Work Transitions in 11 European Countries.Acta Sociologica50 (3): 189210.
Wolter, Stefan C., and Paul Ryan. 2011. Apprenticeship. In , edited by Stephen Machin Eric A.
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
33/36
Appendix A
Figure A.1 The Hungarian compulsory education system
HUNGARY 2009/2010
level ISCED 0 ISCED 1 ISCED 2 ISCED 3
1st cycle 2nd cycle
Grade 1 2 3 4 5 6 7 8 9 10 11 12 13
kindergarten general school academic secondary school prog. ISCED 3a
+
vocational secondary school prog. ISCED 3a
(technikum)
+
vocational training prog. ISCED 3c ++
Age 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
compulsory education until the age of 18 applies for the 1st graders in 1998 and later (previously and from September 2012: until the age of
16)
vocational secondary school programs curriculum includes vocational subjects and many students progress to PS voc to get a VQ
+ : some schools offer an extra grade teaching a foreign language before secondary school educ. (i.e. between grade 8 and 9)
++: some programs are also available for elementary school drop-outs
ISCED English national language share
0 kindergarten voda
1,2a general school ltalnos iskola 100%
3a academic secondary school prog. gimnzium
3a vocational secondary school prog. szakkzpiskola
3c vocational training prog. szakiskola
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
34/36
Table A1: Old and new categories of the national training register (OKJ)
New categories (industries) Original categories in the national training register
Social Services HealthSocial services
Education
Art, culture, communication
Mechanics EngineeringElectrical-engineering, electronics
Informatics
Industry Chemical industry
ArchitectureLight industry
Wood industry
Printing industry
Transportation-environment TransportationEnvironment and water-management
Services Business and economicsManagement
Trade, marketing and administration
Catering, tourismOther Services
Agriculture AgricultureFood industry
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
35/36
35
Table A2: Types of training and employer firms and the number of individuals
1 2 3 4 5 6 7 8 9 1
0
1
1
1
2
1
3
1
4
15 1
6
1
7
1
8
19 2
0
2
1
2
2
2
3
mis total
5thwavetr
ainer
1 Agriculture, forestry and fishing 1 0 1 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 12 17
2 Mining and quarrying 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
3 Processing 6 0 29 2 0 2 1 0 2 0 0 0 0 0 1 1 0 0 2 0 0 0 0 56 102
4 Electricity, gas, steam and air conditioning 1 0 0 7 1 2 0 1 0 0 0 0 0 0 2 0 0 0 1 0 0 0 1 14 30
5 Water supply, wastewater collection and
treatment, waste management
1 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 5
6 Construction 10 0 17 1 1 45 1 1 1 1 0 0 1 1 4 0 1 0 5 0 0 0 0 81 171
7 Trade, automotive services 1 0 15 0 0 4 20 2 7 0 0 0 0 0 3 0 0 0 8 0 0 0 0 70 130
8 Transportation, warehousing 0 0 3 0 0 0 3 1 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 7 16
9 Hotels and restaurants, catreing 2 0 10 0 0 1 17 0 64 0 0 0 0 0 0 0 1 2 5 0 0 0 0 129 231
10 Information, communication
11 Financial and insurance activities 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 4 4
12 Real estate transactions
13 Professional, scientific and technical activities 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1
14 Administrative and support service activities 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 2
15 Administration and defense, compulsory social
security
0 0 0 0 0 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 1 4
16 Education 0 0 3 0 0 0 2 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 9 15
17 Human health and social work 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 21 24
18 Arts, entertainment and recreation 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 1 3
19 Other services 2 0 7 0 3 3 5 2 8 0 0 0 0 1 3 0 1 0 19 0 0 0 1 70 125
20 Households as employers, producers, and service 0 0 1 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3
21 Organizations outside Hungary
22 Other
23 Don't know 0 0 0 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 6 8
missing 21
1
2 44
3
2
4
4
6
15
4
34
5
7
1
21
3
3
3
4
0
1 1
2
2
5
15
4
2
3
5
4
5
8
22
6
1
1
1 4 1
2
6,9
67
9,13
total 23
5
2 53
3
3
4
5
2
21
3
39
5
7
8
29
6
3
4
4
0
1 1
3
2
7
17
3
2
5
5
9
6
0
26
9
1
1
1 4 1
4
7,4
53
10,0
22
-
8/12/2019 Workplace-Based Vocational Training in Central and Eastern Europe - Evidence on the School to Work Transition of Hungarian Apprentices
36/36